The site-specific introduction of non natural amino acids (nnAAs) into a target protein provides a significant advantage for the generation of functionalized protein conjugates over non specific methods (Wang et al., 2011). A variety of non natural amino acids are available that contain moieties that provide bioorthogonal sites for conjugation chemistry and enable specific reactions to occur at these sites. Control over the positions of the conjugation site enables products with optimal function by avoiding active sites and essential protein functional domains. In addition, this allows for the generation of a homogeneous and predictably modified product that improves the functional characteristics and purification of the product.
Site specific incorporation of nnAAs in bacterial cells has been achieved through amino acid substitution approaches and through the engineering of orthogonal aminoacyl tRNA synthetases that charge only their cognate tRNAs with a non natural amino acid. The position of the non natural amino acid in the target protein can be specified by a variety of codons within the gene sequence, but most often it is directed to amber codons. The variety of proteins that can be expressed in E. coli and other prokaryotic based systems, however, is limited by the protein folding machinery of these organisms. Eukaryotic expression systems (such as mammalian expression systems) are capable of expressing a wider variety of proteins including those that require glycosylation for optimal therapeutic function (e.g. G-CSF, insulin, epoietin alpha) exist as protein complexes (e.g. antibodies), or require posttranslational modifications such as disulfide bond formation (e.g. atrial natriuretic factor) that are not accessible in bacterial systems.
Systems for the introduction of nnAAs into mammalian cells have been developed either through transfection of in vitro charged tRNAs (Hecht et al., 1978; Kohrer et al., 2001; Kohrer et al., 2003) or genetically encoded using orthogonal aminoacyl tRNA synthetase/tRNA pairs (Mukai et al 2008, Liu W. et Al., 2007; Wang W., 2007; Ye, S. et Al., 2008; Sakamoto, K et Al., 2002; Takimoto, J. et Al., 2009; Chen, P. et Al., 2009). Chemically acylated tRNAs are not reacylated and thus their use is prohibitive to large scale protein synthesis whether in vitro or in vivo. Genomically encoded RS/tRNA pairs are required to be orthogonal to the host cell in order to retain the specificity of nnAA insertion.
In use of orthogonal aminoacyl tRNA synthetase/tRNA pairs, orthogonality of the RS and tRNA is achieved through mutations at key sites to enable specificity for a nnAA while at the same time reducing or eliminating recognition of canonical amino acids, and host tRNAs. The tRNA may also be modified to prevent cognition by host RSs and to recognize amber stop codons. Several RS/tRNA pairs have been developed including the E. coli TyrRS/B. stearothermophilus tRNAtyr (Liu, W., 2007; Ye et al., 2008; Sakamoto et al., 2002) and E. coli TyrRS/E. coli tRNAtyr (Wang, W., 2007; Takimoto et al., 2009).
It has been observed that one orthogonal RS/tRNA pair naturally evolved in a subset of archaebacteria (methanogenic archaea bacteria that catabolize methylamines) which has specificity for the amino acid pyrrolysine. Pyrrolysine uses a 21st aminoacyl-tRNA synthetase, naturally evolved to be orthogonal to all other amino acids and tRNAs.
Pyrrolysine is a natural amino acid, the only one that is authentically specified by an amber codon. Blight et al., 2004 showed that PylRS and tRNApyl can incorporate Pyrrolysine at amber codons in E. coli. They also showed that the wild type (“WT”) PyLRS is naturally promiscuous and can incorporate analogs of Lysine.
Yokoyama et al (EP1911840) demonstrated that the PylRS/tRNA system is orthogonal in eukaryotic cells and showed the incorporation of several nnAAs into a target proteins encoded by amber codons in bacterial cells. These authors also identified key amino acid residues in pylRS that form the amino acid binding pocket and function in selecting pyrrolysine over other canonical amino acids. Mutations at this site generated mutants able the recognize and aminoacylate the tRNApy with AzZ-lys (Yanagisawa 2008)
This orthogonality extends to bacteria and eukaryotic cells.
PylRS is a naturally promiscuous synthetase that has naturally evolved to exclude lysine, but will incorporate lysine analogs without mutation including azides, alkynes and alkenes, (Yanagisawa et al, 2008; Neumann et al. 2008; Mukai et al., 2008; Nguyen et al., 2009). The basis of this specificity is dependent on hydrophobic interactions between amino acid residues of the pylRS binding pocket with the pyrrol ring of pyrrolysine that stabilizes and correctly positions the amino acid in the active site of the synthetase (Kavran et al., 2007). This RS/tRNA pair has been introduced via transient transfection into bacterial, yeast and mammalian cells and shown to be effective for incorporation of a number of non-natural amino acids into target proteins.
For instance, EP 1911840 demonstrates incorporation of N-ε-boc-Lysine into a target protein in E. coli cells.
The expression of tRNA in eukaryotic cells requires two internal promoters within the tRNA coding sequence. The consensus sequences of such promoters are known as the A-Box and B-Box (Naykova et al., 2003).
Although certain prokaryotic-derived tRNAs naturally carry sequences that function as an internal promoter and can be expressed in animal cells without modifications, or with changes that generate an intragenic promoter sequence but do not alter the function of the tRNA or its recognition by its cognate RS, tRNAPyl does not contain such promoter. Furthermore, the D loop where A-Box and B-Box are normally present is unusually small and the introduction of said A-Box and B-Box would destroy its function as reported in yeast by Hancock et al (2010) and confirmed by the inventors in mammalian cells.
WO2007099854 describes the use of a eukaryotic snRNA promoter to drive tRNAPyl expression in eukaryotic cells. DNA constructs described therein comprise the tRNApyl gene, a transcription terminator sequence placed 3′ of said tRNA gene, and a promoter sequence that induces transcription by RNA Polymerase II or III such as U1 snRNA promoter or U6 snRNA promoter placed 5′ to said tRNApyl gene.
Mammalian expression is of particular interest as it allows for the production of fully folded proteins and protein complexes like full length antibodies that are challenging or currently inaccessible to prokaryotic systems or yeast cells.
Transient transfection experiments of genes encoding the pyrrolysine aminoacyl tRNA synthetase (pylRS) and its tRNApyl, in both human (HEK293) and hamster (CHO) cells, have shown that the pylRS/tRNA pair efficiently incorporates nnAAs into a target protein at sites designated by an amber stop codon in mammalian cells (see for instance Mukai 2008).
EP1911840 teaches the introduction of a vector carrying a WT PylRS, a vector carrying a tRNApyl gene, and a vector carrying a target gene where an amber mutation is introduced at the site where the lysine derivative is to be inserted. The only technique utilized to introduce the vectors is transient transfection. In fact, nowhere in the patent application the selection of stable clones is mentioned nor applied experimentally.
WO09038195 describes the generation of mutant PylRS enzymes in order to improve its catalytic activity, and allow incorporation of non natural amino acids derived from lysine with bulky side chain structures.
In particular, WO09038195 describes a mutation at position 384 (referred to methanosarcina mazei PylRS amino acid sequence) whereby Tyr384 is replaced with Phenylalanine, among other amino acids. It is hypothesized that due to the fact that Tyr384 interacts with a substrate amino acid, particularly with its main chain (Kavran 2007, Nozawa 2009) there is likelihood that the enzyme catalytic activity would be enhanced independently of the substrate.
As noted above, expression based on the PylRS and tRNApyl orthogonal pair has hitherto only been achieved in transiently stable eukaryotic cell lines. Transiently stable cell lines are not suitable for the reliable manufacture of commercial products; indeed the present inventors believe that the biologic products on the market today derived from mammalian cells are exclusively generated by stable cell lines.
Therefore there remains a need in the art to develop methods for the production of stable eukaryotic cells containing the PylRS and tRNApyl orthogonal pair thereby to facilitate production of proteins containing nnAAs on a commercial scale.
The present invention addresses the aforementioned need.
Pyrrolysine analogs, defined as amino acid derivatives recognized by either native or genetically evolved PylRS and incorporated into proteins at amber codon sites, have been disclosed in the past few years and reviewed, for instance, by Feckner et. al (Fekner, Li, & Chan, 2010) and Liu et al. Analogs bearing functional groups or post translational modifications have been site-specifically incorporated into proteins using pylRS-tRNApyl systems. Several studies, see e.g. Yanagisawa et al, focused on mutations within the PylRS enzyme in order to accommodate analogs in which the N6 substituent were an aromatic ring within the binding pocket pyrrolysine. Others, for instance Nguyen et al (also in WO2010/139948), and Li et al (also in WO2011/044255) focused on identification of pyrrolysine analogs which do not carry a bulky N6 substituent, with the result of obtaining simpler analogs which would be simple to synthesize and interact with native pylRS/tRNApyl pairs. There remains a need to develop further pyrrolysine analogs. Whilst pyrrolysine analogs made thus far have been restricted to those evolved from a lysine backbone, the present inventors have generated pyrrolysine analogs successfully incorporated into proteins with native pylRS/tRNApyl pairs starting from a variety of amino acid structures.
Antibody drug conjugates (ADCs), composed of recombinant chimeric, humanized or human antibodies covalently bound by means of synthetic linkers to highly cytotoxic drugs, have been developed in recent years in order to target cytotoxic drugs to tumor cells. The right combination of antibodies targeting tumor associated antigens, a potent toxins and appropriate conjugation chemistry can be very effective at delivering the toxin directly to the tumor cells, while avoiding toxicity of the drug to normal tissue.
ADCs developed so far are heterogeneous mixtures of conjugated and unconjugated antibody, depending on the chemistry of the conjugation used when generating the ADC. In particular, the random nature of the most commonly used conjugation protocols results in a collection of species with varying numbers of drugs conjugated per antibody molecule (DAR) as well as varying conjugation sites. Common conjugation chemistries include lysine side-chain based conjugation, which results in a wide range of species due to the large availability of lysine residues in a typical antibody. More site-specific conjugations have been obtained through engineering of cysteine residues to produce reactive thiol groups, resulting in nearly homogeneous ADCs.
Her2 tumor associated antigen, a member of the EGFR family, has been successfully targeted in breast cancer with Herceptin, an anti-Her 2 antibody, however, the antibody itself is effective in a limited group of patients. A more potent form, ado-trastuzumab emtansine, which has a toxin linked to it, is now available. Ado-trastuzumab emtansine is able to effectively treat patients who are refractory to Herceptin, due to the ability of ado-trastuzumab emtansine to deliver a toxin to the cytoplasm of the cancer cell. The conjugation chemistry being used by ado-trastuzumab emtansine and Brentuximab vedotin, exploits existing cysteine residues that normally form disulfide bonds, and more recently, engineered free cysteine residues. This approach has led to the production of heterogeneous mixtures of ADC with different numbers of drug at different positions on the mAb. The linkers used include thioether (Kadcyla) as well as dipeptide linkers (Adcetris), the latter being specifically cleaved by lysosomal acid hydrolases. Both types of linkers appear to be effective, but not optimal. Conventional Cys or Lys directed bioconjugation methods such as those used for manufacture of by ado-trastuzumab emtansine, Brentuximab vedotin, and gemtuzumab ozogamicin permit premature release of toxin prior to tumor cell engagement. Gemtuzumab ozogamicin, which was approved in 2000, was withdrawn from the market in 2010, due to high toxicity due to the use of an acid labile linker which caused intolerable release of the toxin from the ADC in the blood.
Therefore there remains a need in the art to develop highly homogeneous ADCs, where the conjugation sites and the number of drugs per antibody are well controlled.
The present inventors have found that through use of site specific incorporation of nnAAs and subsequent conjugation of antibodies at the site of nnAA it is possible to generate homogeneous and potent ADCs. Furthermore, the present inventors have found sites, within the IgG constant region, which can be used for conjugation without disrupting the specificity of the binding of the antibody or its pharmacokinetic properties in vivo.